
import pandas as pd
import numpy as np
from tqdm import tqdm
from itertools import product

## Agregar las librerias que iran a usar
from sklearn.preprocessing import StandardScaler
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.decomposition import PCA
from pca import pca

#Clustering
from sklearn.cluster import KMeans
from fcmeans import FCM
from sklearn.mixture import GaussianMixture
from sklearn.cluster import DBSCAN
from sklearn.neighbors import NearestNeighbors

#Outliers
from sklearn.neighbors import LocalOutlierFactor
from sklearn.ensemble import IsolationForest
from sklearn.svm import OneClassSVM

#Metrics
from sklearn.metrics import silhouette_score, davies_bouldin_score

#Librerias para trabajar texto
from string import punctuation
from unidecode import unidecode

# Mute all future warnings
import warnings
warnings.filterwarnings("ignore")


dataset = pd.read_csv("musics.csv")
dataset


dataset.shape

(447622, 36)


dataset.columns

Index(['artist_name', 'artist_id', 'album_id', 'album_type',
       'album_release_date', 'album_release_year',
       'album_release_date_precision', 'danceability', 'energy', 'key',
       'loudness', 'mode', 'speechiness', 'acousticness', 'instrumentalness',
       'liveness', 'valence', 'tempo', 'track_id', 'analysis_url',
       'time_signature', 'disc_number', 'duration_ms', 'explicit',
       'track_href', 'is_local', 'track_name', 'track_preview_url',
       'track_number', 'type', 'track_uri', 'external_urls.spotify',
       'album_name', 'key_name', 'mode_name', 'key_mode'],
      dtype='object')


dataset.loc[191428]

artist_name                                                           Linkin Park
artist_id                                                  6XyY86QOPPrYVGvF9ch6wz
album_id                                                   6PFPjumGRpZnBzqnDci6qJ
album_type                                                                  album
album_release_date                                                     2000-10-24
album_release_year                                                         2000.0
album_release_date_precision                                                  day
danceability                                                                0.556
energy                                                                      0.864
key                                                                             3
loudness                                                                    -5.87
mode                                                                            0
speechiness                                                                0.0584
acousticness                                                              0.00958
instrumentalness                                                              0.0
liveness                                                                    0.209
valence                                                                       0.4
tempo                                                                     105.143
track_id                                                   7q115ia4fQn9zonjpexWsY
analysis_url                    https://api.spotify.com/v1/audio-analysis/7q11...
time_signature                                                                  4
disc_number                                                                     1
duration_ms                                                                216880
explicit                                                                    False
track_href                      https://api.spotify.com/v1/tracks/7q115ia4fQn9...
is_local                                                                    False
track_name                                                             In the End
track_preview_url               https://p.scdn.co/mp3-preview/6ce8bcf317e8c562...
track_number                                                                    8
type                                                                        track
track_uri                                    spotify:track:7q115ia4fQn9zonjpexWsY
external_urls.spotify           https://open.spotify.com/track/7q115ia4fQn9zon...
album_name                                                          Hybrid Theory
key_name                                                                       D#
mode_name                                                                   minor
key_mode                                                                 D# minor
Name: 191428, dtype: object


## Seleccion de algunas columnas (features)
col_features = ['acousticness', 'danceability', 'energy', 'instrumentalness', 'liveness', 
               'loudness','speechiness','valence','tempo', 'key', 
               'mode', 'time_signature', 'duration_ms']

## Mostrar algunas estadisticas
dataset[col_features].describe()


#Subconjunto canciones del año 2000

dataset_sub = dataset[dataset['album_release_year']==2000]
dataset_sub.shape

(4103, 36)


#Limpieza de duplicados

#Primero debemos trabajar los titulos de las canciones para poder identificarlos lo mejor posible.

#Limpiamos espacios extras
dataset_sub.loc[:,['artist_name','track_name']] = dataset_sub[['artist_name','track_name']].apply(lambda x: x.str.strip())

#Limpiamos caracteres especiales y puntuaciones
def strip_words(x):
    x = str(x)
    non_words = list(punctuation)
    sentece = ''.join([letra for letra in x if letra not in non_words]) #Quitamos carateres y puntuaciones raras
    new_sentence = unidecode(sentece)
    return new_sentence

dataset_sub.loc[:,['artist_name','track_name']] = dataset_sub[['artist_name','track_name']].applymap(strip_words)

#Dejamos todo en minusculas
dataset_sub.loc[:,['artist_name','track_name']] = dataset_sub[['artist_name','track_name']].apply(lambda x: x.str.title())

#Chequeamos duplicados
print(f"Nº de canciones duplicadas: {dataset_sub.duplicated(subset=['artist_id','track_name']).sum()}")


#Existen canciones duplicadas. Dicho esto trataremos de dejar solamente las que no son en vivo (para un mejor sonido) y y las mas recientes.
dataset_sub = dataset_sub.sort_values(by=['artist_name','track_name','liveness','album_release_year'], ascending=[True,True,True,False])

#Luego de ordenar la data, nos aseguramos de quedarnos con el primer registro.
dataset_sub = dataset_sub.drop_duplicates(subset=['artist_name','track_name'],keep='first')

Nº de canciones duplicadas: 959


#Selección de features recomendadas
data = dataset_sub[col_features]

#Adicionalmente se obtienen las columnas artist_name, album_name, track_name y track_id para poder crear nuestro playlist mas adelante
data_info = dataset_sub[['artist_name','album_name','track_name','track_id']]

print(data.shape,data_info.shape)

(3144, 13) (3144, 4)


#Chequeamos si existen datos faltantes
print(data.isna().sum(), data_info.isna().sum())

#No existen datos faltantes

acousticness        0
danceability        0
energy              0
instrumentalness    0
liveness            0
loudness            0
speechiness         0
valence             0
tempo               0
key                 0
mode                0
time_signature      0
duration_ms         0
dtype: int64 artist_name    0
album_name     0
track_name     0
track_id       0
dtype: int64


#Observamos si existe correlacion entre variables. Si existe alguna con una correlacion >0.75, tendremos que dropear una de las 2.

plt.figure(figsize=(15,10))
sns.heatmap(data=data.corr(), annot=True, fmt=".2f", cmap="YlGnBu")
plt.show()


#La variable loudness y energy tienen una alta correlacion. Dropearemos loudness.

data = data.drop('loudness', axis=1)
data.head()


#Dada la diferencia de magnitud de los datos, se debe normalizar los datos, de lo contrario nuestros modelos no nos daran buenos resultados.

scaler = StandardScaler()
data_scaled = pd.DataFrame(scaler.fit_transform(data),columns=data.columns, index=data.index)
data_scaled


#Veremos si existen outliers (Al menos usando boxplots).

data_scaled.boxplot()
plt.xticks(rotation=45,ha='right')
plt.title('data sin reduccion de outliers')
plt.show()


#Usando el metodo de 3 desvaciones estandar. Veremos cuantas filas contienen a lo menos 1 outlier.

outliers = (data_scaled>3 )|(data_scaled<-3)

print(f"Porcentaje de filas con outliers con metodo de 3 desvaciones estandar: {outliers.any(axis=1).mean():.2%}")

#Graficando
model = PCA(n_components=2).fit(data_scaled)
x_pca = model.transform(data_scaled)
sns.scatterplot(x = x_pca[:,0], y = x_pca[:,1], hue = outliers.any(axis=1), palette = ['#33658a','#fe4a49'])
plt.title('outliers con 3 desviaciones estandar')
plt.show()

Porcentaje de filas con outliers con metodo de 3 desvaciones estandar: 8.43%


# Ahora este ultimo metodo no seria ten eficiente si no tenemos distribuciones normales,lo que justamente aqui no pasa para todas las variables.
plt.rcdefaults()
data_scaled.hist(grid=False)
plt.tight_layout()
plt.show()


# Acorde a lo anterior tenemos que 8.43% de las filas tienen al menos 1 outlier, pero sabemos que este metodo no toma en cuenta la interacción entre variables.
# Dicho esto usaremos SVM para detectar outliers, usando como proxy de contaminacion el 8.43% del modelo anterior.

outliers = OneClassSVM(nu=0.0843, kernel = 'rbf', gamma = 'auto',max_iter=-1).fit_predict(data_scaled)

print(f"Porcentaje de filas con outliers con medelo SVM - kernel rbf: {(outliers==-1).sum()/data_scaled.shape[0]:.2%}")

#Graficando
model = PCA(n_components=2).fit(data_scaled)
x_pca = model.transform(data_scaled)
sns.scatterplot(x = x_pca[:,0], y = x_pca[:,1], hue = [True if i == -1 else False for i in outliers], palette = ['#33658a','#fe4a49'])
plt.title('outliers con SVM - kernel rbf')
plt.show()

Porcentaje de filas con outliers con medelo SVM - kernel rbf: 8.49%


#Con la intencion de mantener consitencia entre ambos datasets, se eliminan los outliers de ambos. Recordemos que vamos a necesitar la info para crear nuestra playlist.

no_outliers = [False if i == -1 else True for i in outliers]

# Nos quedamos con nuestras filas sin outliers para ambos datasets

data_scaled = data_scaled[no_outliers]
data_info = data_info[no_outliers]


#Visualizamos con reduccion de outliers

data_scaled.boxplot()
plt.xticks(rotation=45,ha='right')
plt.title('data con reduccion de outliers')
plt.show()


data_scaled.shape

(2877, 12)


#Buscando el mejor modelo

#Se recomienda usar max cluser = numero de atributos + 1
max_n_clusters = data_scaled.shape[1]+1

## Almacenado de errores al cuadrado
sse = []

for k in range(1, max_n_clusters):

    ## Instancia del modelo
    kmeans_model = KMeans(n_clusters=k,n_init='auto', init='k-means++',random_state=0, max_iter=300, tol=0.0001)

    ## Ajuste del modelo
    kmeans_model.fit(data_scaled)

    ## Guardar el SSE obtenido
    sse.append(kmeans_model.inertia_)

plt.figure(figsize=(10, 5))
plt.plot(range(1, max_n_clusters), sse)
plt.title('Elbow curve')
plt.xlabel('Numero de cluster')
plt.ylabel('SSE')
plt.tight_layout()
plt.show()


max_n_clusters = data_scaled.shape[1]+1

## Almacenado de SSE-score
sse = []
silueta_km_lst = []
davies_bouldin_km_lst = []
n_cluster_lst = []

for k in range(2, max_n_clusters):

    ## Instancia del modelo
    kmeans_model = KMeans(n_clusters=k,n_init='auto', init='k-means++',random_state=0, max_iter=300, tol=0.0001)

    ## Ajuste del modelo
    kmeans_model.fit(data_scaled)


    # Cálculo de métricas para K-Means
    silueta_km = silhouette_score(data_scaled, kmeans_model.labels_)
    davies_bouldin_km = davies_bouldin_score(data_scaled, kmeans_model.labels_)
    silueta_km_lst.append(silueta_km)
    davies_bouldin_km_lst.append(davies_bouldin_km)
    n_cluster_lst.append(k)


silueta_davis = pd.DataFrame()

silueta_davis['n_clusters'] = n_cluster_lst
silueta_davis['silueta']=silueta_km_lst 
silueta_davis['davies_bouldin']=davies_bouldin_km_lst
print(silueta_davis)

fig, ax = plt.subplots(figsize=(10, 5))
y2 = ax.twinx()

sns.lineplot(data=silueta_davis, x='n_clusters', y='silueta', ax=ax,label='Silueta', color='blue')
sns.lineplot(data=silueta_davis, x='n_clusters', y='davies_bouldin', ax=y2, label='Davies Bouldin', color='red')

# Add labels at the end of each line
silueta_max = last_row_silueta = silueta_davis.iloc[-1]['silueta']
davies_max = last_row_silueta = silueta_davis.iloc[-1]['davies_bouldin']

ax.text(silueta_davis['n_clusters'].iloc[-1], silueta_max, 'Silueta', color='blue', va='center')
y2.text(silueta_davis['n_clusters'].iloc[-1], davies_max, 'Davies Bouldin', color='red', va='center')

# Set labels for both x-axes
ax.set_xlabel('Number of Clusters')
y2.set_xlabel('Number of Clusters')
ax.legend_.remove()
y2.legend_.remove()
plt.xlim(1, 14)
plt.title('Score de silueta y davies_bouldin')
plt.show()

    n_clusters   silueta  davies_bouldin
0            2  0.250665        1.551570
1            3  0.183340        2.027633
2            4  0.152244        2.120622
3            5  0.161261        1.949095
4            6  0.162803        1.894481
5            7  0.145194        1.966472
6            8  0.155780        1.949053
7            9  0.134181        1.877869
8           10  0.135119        1.801033
9           11  0.138361        1.769663
10          12  0.133374        1.740106


#Modelo con mejor parametro

kmeans_model = KMeans(n_clusters=2,n_init='auto', init='k-means++',random_state=0, max_iter=300, tol=0.0001)

## Ajuste del modelo con los datos
kmeans_model.fit(data_scaled)

## Transformación de los datos usando PCA
pca_model = PCA(n_components=2, random_state=0)
pca_model.fit(data_scaled)
pca_data = pca_model.transform(data_scaled)

## Crear un dataframe con los datos transformados mediante PCA y agregar las etiquetas de los clusteres
pca_data = pd.DataFrame(pca_data, columns=['PC1', 'PC2'])
pca_data['cluster'] = kmeans_model.predict(data_scaled)

## Graficar los datos por clases. 
N = pca_data['cluster'].nunique()

plt.figure(figsize=(8, 6))
sns.scatterplot(data=pca_data, 
                x='PC1', y='PC2', 
                hue='cluster',
                palette=sns.color_palette()[:N])
plt.legend(loc=[1.01, 0.5], title='Cluster')
plt.title('Cluster formados')
plt.show()


#Varianza explicada por cantidad de PCA

pca_model = PCA()
pca_model.fit_transform(data_scaled)
pca_var = pd.DataFrame([pca_model.explained_variance_ratio_,
              pca_model.explained_variance_ratio_.cumsum()]).T \
            .rename(columns={0:'variance',1:'cumulative_variance'})
pca_var['N_PCA']  = [i+1 for i in range (len(pca_var.variance))]
pca_var


#Para poder visualizar los datos en un espacio 2D, usaremos las primeras 2 componentes.
model = pca(verbose=False,n_components=2,random_state=0)

#En la siguiente tabla se observa que features contribuyen mas a la varianza de las primeras 2 componentes.
model.fit_transform(data_scaled)['topfeat']


#Ahora iremos harmando nuestro biplot, para ver que features y que magnitud corresponde a cada cluster.

#Primero nuestro scatterplot con la etiquetas respectivas para nuestras PC1 y PC2
model.scatter(labels=kmeans_model.labels_, legend=True,figsize=(15,8))
plt.show()

[scatterd] >INFO> Create scatterplot


# 2D plot with samples colored on color_intensity
model.biplot(labels=kmeans_model.labels_, legend=True, cmap=None,figsize=(15,8))
plt.show()

[scatterd] >INFO> Create scatterplot


# 2D plot with samples colored on color_intensity
model.biplot(labels=kmeans_model.labels_, legend=True,alpha=0.4,figsize=(15,8),cmap='Pastel2')
plt.show()

[scatterd] >INFO> Create scatterplot


data_scaled['cluster'] = kmeans_model.labels_
data_melt = data_scaled.melt(id_vars=['cluster'])

plt.rcdefaults()

# Create a figure with subplots
fig, axes = plt.subplots(1,len(data_melt['cluster'].unique()), figsize=(12, 5))

# Enumerate through unique cluster values
for i, col in enumerate(data_melt['cluster'].unique()):
    # Select the axis for the current subplot
    ax = axes[i]
    
    # Filter the data for the current cluster
    subset_data = data_melt[data_melt['cluster'] == col]
    
    # Create a point plot on the current axis
    sns.boxplot(data=subset_data, x='variable', y='value', ax=ax,showfliers=False)
    
    # Customize the x-axis labels (rotation, alignment)
    ax.set_xticklabels(ax.get_xticklabels(), rotation=45, ha='right',fontsize=10)
    # Add a title to the subplot
    ax.set_title(f'Cluster {col}')
    ax.set_xlabel('')
    
# Adjust spacing between subplots
plt.tight_layout()

# Show all subplots together
plt.show()


#Volvemos nuestros datos al estado original para evaluar en base a las documentacion de spotify
df_detail = pd.DataFrame(scaler.inverse_transform(data_scaled.drop(columns='cluster')),columns=data_scaled.drop(columns='cluster').columns,index=data_scaled.index)
df_detail['cluster'] = kmeans_model.labels_


df_detail.groupby('cluster').mean().reset_index()


df = pd.concat([data_info,df_detail],axis=1).reset_index(drop=True)
df.head(5)


# Uno de mis grupos favoritos es Linkin Park. Por lo que seleccionare el cluster de esta banda para compararlos con otras que sean del mismo.

df[df.artist_name=='Linkin Park'].head()


#Seleccionamos la cancion Papercut y extraemos el cluster

cluster = df[(df.artist_name=='Linkin Park')&(df.track_name=='Papercut')].iloc[0,-1]
cluster

0


#Artistas dentro del mismo cluster

df_cluster = df[df.cluster==cluster]
df_cluster.artist_name.unique()

array(['2Pac', 'Acdc', 'Alejandro Sanz', 'Alex Bueno', 'Andres Calamaro',
       'Aterciopelados', 'Attaque 77', 'Babasonicos', 'Backstreet Boys',
       'Belle  Sebastian', 'Ben Harper', 'Bersuit Vergarabat', 'Bjork',
       'Black Eyed Peas', 'Bon Jovi', 'Caetano Veloso',
       'Camille Saintsaens', 'Chancho En Piedra', 'Chayanne',
       'Children Of Bodom', 'Coldplay', 'Collective Soul',
       'Daniela Mercury', 'David Arkenstone', 'Deep Purple', 'Deftones',
       'Diego Torres', 'Divididos', 'Elton John', 'Eminem', 'Erasure',
       'Eric Clapton', 'Fatboy Slim', 'Fito Paez', 'Fleetwood Mac',
       'Franco De Vita', 'Gilberto Gil', 'Gondwana', 'Green Day',
       'Hans Zimmer', 'Helloween', 'Iggy Pop', 'In Flames',
       'Intiillimani', 'Ismael Serrano', 'Joan Manuel Serrat',
       'John Coltrane', 'Jose Luis Perales', 'Juanes', 'Julieta Venegas',
       'Kamelot', 'Kevin Johansen', 'La Mosca Tsetse',
       'La Oreja De Van Gogh', 'La Renga', 'Limp Bizkit', 'Linkin Park',
       'Los Angeles Azules', 'Los Autenticos Decadentes',
       'Los Enanitos Verdes', 'Los Jaivas', 'Los Miserables',
       'Los Pericos', 'Los Piojos', 'Los Terricolas', 'Los Tres Ases',
       'Lucybell', 'Luis Fonsi', 'Luis Miguel', 'Lynyrd Skynyrd',
       'Marco Antonio Solis', 'Matchbox Twenty', 'Matt Uelmen',
       'Miguel Bose', 'Millencolin', 'Moby', 'Nacao Zumbi',
       'Nana Mouskouri', 'Neil Young', 'Nightwish', 'Oasis', 'Outkast',
       'Pantera', 'Papa Roach', 'Paul Mccartney', 'Pearl Jam',
       'Pyotr Ilyich Tchaikovsky', 'Queens Of The Stone Age',
       'Quilapayun', 'Radiohead', 'Rage Against The Machine',
       'Reo Speedwagon', 'Rhapsody', 'Ricardo Arjona', 'Ricky Martin',
       'Sergei Rachmaninoff', 'Sergio Mendes', 'Shakira', 'Sr71',
       'Sui Generis', 'Sum 41', 'The Avalanches', 'The Black Keys',
       'The Cure', 'The Hives', 'The Offspring', 'The White Stripes',
       'Therion', 'Thievery Corporation', 'Tiesto', 'Tommy Guerrero',
       'Tracy Chapman', 'U2', 'Vicente Fernandez', 'Victor Manuelle',
       'Wisin  Yandel', 'Wyclef Jean'], dtype=object)


#Creando playlist de 3 horas
playlist = pd.DataFrame(columns=df_cluster.columns)

while True:

    if playlist.duration_ms.sum() >= 1.08e+7: #3 horas en milisegundos
        break
    else:
        song = df_cluster.sample(1)
        if song.track_id.item() not in playlist.track_id.values:
            playlist = pd.concat([playlist, song])
        else:
            continue

#Nº Canciones en la playlist
print(f'Canciones totales en la playlist: {playlist.shape[0]}')

#Duracion en horas
print(f'Duración en horas: {playlist.duration_ms.sum()/1000/60/60:.2f}')

Canciones totales en la playlist: 44
Duración en horas: 3.03


playlist[['artist_name','track_name']]


import spotipy
from spotipy.oauth2 import SpotifyClientCredentials


#Conectandose con spotify

from spotify_keys import cid, secret

cid = cid
secret = secret

client_credentials_manager = SpotifyClientCredentials(client_id=cid, client_secret=secret)
sp = spotipy.Spotify(client_credentials_manager=client_credentials_manager)


#Top 20 canciones de polyphia
results = sp.search(q='polyphia', limit=20)
for idx, track in enumerate(results['tracks']['items']):
    print(idx, track['name'])

0 Playing God
1 G.O.A.T.
2 Ego Death (feat. Steve Vai)
3 ABC (feat. Sophia Black)
4 40oz
5 Champagne
6 Ego Death (feat. Steve Vai)
7 O.D.
8 The Worst
9 Neurotica
10 Playing God
11 Reverie
12 Chimera (feat. Lil West)
13 Euphoria
14 Goose
15 Bloodbath (feat. Chino Moreno)
16 ABC (feat. Sophia Black)
17 So Strange (feat. Cuco)
18 All Falls Apart
19 Genesis (feat. Brasstracks)


#Obtenemos sus features
features_lst=[]

for idx, track in enumerate(results['tracks']['items']):
    trackID = track['uri']
    features = sp.audio_features(trackID)[0]
    features['artist_name'] = track['artists'][0]['name']
    features['track_name'] = track['name']
    features_lst.append(features)

spotify_df = pd.DataFrame(features_lst)


#Recordemos que dropeamos la variables loudness
col_features = ['acousticness',
 'danceability',
 'energy',
 'instrumentalness',
 'liveness',
 'speechiness',
 'valence',
 'tempo',
 'key',
 'mode',
 'time_signature',
 'duration_ms']

#Escalamos los datos antes de entregarlos al modelo
spotify_scaled = pd.DataFrame(scaler.fit_transform(spotify_df[col_features]),columns=col_features)

#Prediciendo el cluster para cada cancion
predicted_values = kmeans_model.predict(spotify_scaled)

#juntamos todo
pd.concat([spotify_df, pd.Series(predicted_values,name='cluster')], axis=1)


from spotipy.oauth2 import SpotifyOAuth

scope = "playlist-modify-public"
sp = spotipy.Spotify(auth_manager=SpotifyOAuth(scope=scope, client_id=cid, client_secret=secret, redirect_uri='http://localhost:3000'))


# sp.me()


user = sp.me()['id']
spotify_playlist = sp.user_playlist_create(user=user, 
                        name='DAD_Playlist')
spotify_playlist_id = spotify_playlist['id']


## ID del playlist creado ubicado en uri
newPlaylistId = spotify_playlist_id


## Extraer los track_id de la lista de canciones seleccionadas
trackIDs = list(playlist.track_id)


## Agregar la lista de canciones al playlist de Spotify
sp.user_playlist_add_tracks(user, newPlaylistId, trackIDs)

{'snapshot_id': 'Miw2ZjM5MjQyNmJhMTk1MDczMDVjMmJhNmU1ZWU2NjJiMWE4YTliNDgz'}


print(f"Lista disponible en: {spotify_playlist['external_urls']['spotify']}")

Lista disponible en: https://open.spotify.com/playlist/74NwjpUzI8VXxi9jsnWVWH

	artist_name	artist_id	album_id	album_type	album_release_date	album_release_year	album_release_date_precision	danceability	energy	key	...	track_name	track_preview_url	track_number	type	track_uri	external_urls.spotify	album_name	key_name	mode_name	key_mode
0	2Pac	1ZwdS5xdxEREPySFridCfh	1nGbXgS6toEOcFCDwEl5R3	album	2019-08-01	2019.0	day	0.656	0.882	0	...	California Love	https://p.scdn.co/mp3-preview/93e456ef0b73f23f...	1	track	spotify:track:6ayeqYtOtwVhqVB6k6MKoh	https://open.spotify.com/track/6ayeqYtOtwVhqVB...	California Love	C	major	C major
1	2Pac	1ZwdS5xdxEREPySFridCfh	1nGbXgS6toEOcFCDwEl5R3	album	2019-08-01	2019.0	day	0.810	0.642	8	...	Slippin' Into Darkness	https://p.scdn.co/mp3-preview/440595604d3f4946...	2	track	spotify:track:1UDsnzBp8gUCFsrzUDlZI9	https://open.spotify.com/track/1UDsnzBp8gUCFsr...	California Love	G#	major	G# major
2	2Pac	1ZwdS5xdxEREPySFridCfh	1nGbXgS6toEOcFCDwEl5R3	album	2019-08-01	2019.0	day	0.548	0.590	4	...	Ride or Die	https://p.scdn.co/mp3-preview/cc18dc90d609d375...	3	track	spotify:track:3bKs15o7F9VP6GBExCbb6H	https://open.spotify.com/track/3bKs15o7F9VP6GB...	California Love	E	minor	E minor
3	2Pac	1ZwdS5xdxEREPySFridCfh	1nGbXgS6toEOcFCDwEl5R3	album	2019-08-01	2019.0	day	0.839	0.657	5	...	I Ain't Mad At Cha	https://p.scdn.co/mp3-preview/d138f0170423cd9a...	4	track	spotify:track:4L0iAst3yLonw8aGxTRCvb	https://open.spotify.com/track/4L0iAst3yLonw8a...	California Love	F	minor	F minor
4	2Pac	1ZwdS5xdxEREPySFridCfh	1nGbXgS6toEOcFCDwEl5R3	album	2019-08-01	2019.0	day	0.854	0.694	0	...	Static II	https://p.scdn.co/mp3-preview/dddb7d0ea0205338...	5	track	spotify:track:66men3J5qFERvIY06M5hQ9	https://open.spotify.com/track/66men3J5qFERvIY...	California Love	C	minor	C minor
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
447617	ZZ Top	2AM4ilv6UzW0uMRuqKtDgN	0Y9jM9umdMOH7t19urnOw4	album	1970-01-16	1970.0	day	0.673	0.520	0	...	Neighbor, Neighbor	https://p.scdn.co/mp3-preview/530e4ce805075d4e...	6	track	spotify:track:4snfGOJDRVfY9jH43CsmJB	https://open.spotify.com/track/4snfGOJDRVfY9jH...	ZZ Top's First Album	C	major	C major
447618	ZZ Top	2AM4ilv6UzW0uMRuqKtDgN	0Y9jM9umdMOH7t19urnOw4	album	1970-01-16	1970.0	day	0.674	0.397	4	...	Certified Blues	https://p.scdn.co/mp3-preview/ee79a42629975bd9...	7	track	spotify:track:7mdPXhJBfxJBonEmqZvm9t	https://open.spotify.com/track/7mdPXhJBfxJBonE...	ZZ Top's First Album	E	minor	E minor
447619	ZZ Top	2AM4ilv6UzW0uMRuqKtDgN	0Y9jM9umdMOH7t19urnOw4	album	1970-01-16	1970.0	day	0.466	0.406	7	...	Bedroom Thang	https://p.scdn.co/mp3-preview/dd0df120d0b10f80...	8	track	spotify:track:23dC4zCpB1bnLzBxAXDLD7	https://open.spotify.com/track/23dC4zCpB1bnLzB...	ZZ Top's First Album	G	major	G major
447620	ZZ Top	2AM4ilv6UzW0uMRuqKtDgN	0Y9jM9umdMOH7t19urnOw4	album	1970-01-16	1970.0	day	0.611	0.337	0	...	Just Got Back from Baby's	https://p.scdn.co/mp3-preview/5fa536adf7c6117e...	9	track	spotify:track:7AX7qVqFljHOJC3FIQ8a6t	https://open.spotify.com/track/7AX7qVqFljHOJC3...	ZZ Top's First Album	C	major	C major
447621	ZZ Top	2AM4ilv6UzW0uMRuqKtDgN	0Y9jM9umdMOH7t19urnOw4	album	1970-01-16	1970.0	day	0.635	0.645	4	...	Backdoor Love Affair	https://p.scdn.co/mp3-preview/311f47b27f09ff6e...	10	track	spotify:track:7w83TSpDeFsL8dV9pAHWn6	https://open.spotify.com/track/7w83TSpDeFsL8dV...	ZZ Top's First Album	E	minor	E minor

	acousticness	danceability	energy	instrumentalness	liveness	loudness	speechiness	valence	tempo	key	mode	time_signature	duration_ms
count	447622.000000	447622.000000	447622.000000	447622.000000	447622.000000	447622.000000	447622.000000	447622.000000	447622.000000	447622.000000	447622.000000	447622.000000	4.476220e+05
mean	0.698733	0.391103	0.340512	0.506068	0.221720	-18.671858	0.068922	0.337392	108.736640	5.061246	0.683420	3.727714	2.291097e+05
std	0.369360	0.183373	0.317557	0.413743	0.219318	8.640242	0.094167	0.274319	31.650459	3.491417	0.465142	0.717955	1.766587e+05
min	0.000000	0.000000	0.000000	0.000000	0.000000	-60.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	1.066000e+03
25%	0.394000	0.252000	0.075600	0.001690	0.096800	-24.445000	0.037800	0.089400	82.391000	2.000000	0.000000	4.000000	1.234400e+05
50%	0.923000	0.370000	0.210000	0.715000	0.123000	-19.476500	0.044300	0.274000	105.723500	5.000000	1.000000	4.000000	1.949610e+05
75%	0.986000	0.514000	0.582000	0.901000	0.253000	-11.644000	0.058400	0.537000	131.051000	8.000000	1.000000	4.000000	2.715600e+05
max	0.996000	0.986000	1.000000	1.000000	1.000000	0.496000	0.971000	0.996000	244.952000	11.000000	1.000000	5.000000	4.796395e+06

	acousticness	danceability	energy	instrumentalness	liveness	speechiness	valence	tempo	key	mode	time_signature	duration_ms
387	0.608	0.506	0.413	0.000000	0.2140	0.2630	0.581	98.720	6	1	4	140373
384	0.177	0.718	0.666	0.000003	0.2150	0.3610	0.643	120.544	11	1	5	182440
385	0.183	0.612	0.704	0.000000	0.1130	0.5600	0.140	82.276	8	0	4	294400
379	0.126	0.603	0.733	0.000008	0.0766	0.4270	0.489	82.902	8	1	4	176866
399	0.433	0.758	0.542	0.000000	0.0857	0.0502	0.591	93.377	1	0	4	199800

	acousticness	danceability	energy	instrumentalness	liveness	speechiness	valence	tempo	key	mode	time_signature	duration_ms
387	0.263767	0.265679	-0.223575	-0.711196	-0.055430	2.223782	0.625767	-0.479719	0.183876	0.738730	0.304075	-0.660752
384	-0.823009	1.355166	0.551399	-0.711187	-0.050841	3.349758	0.848299	0.222255	1.660895	0.738730	2.167648	-0.469731
385	-0.807880	0.810422	0.667798	-0.711196	-0.518883	5.636179	-0.957086	-1.008645	0.774683	-1.353674	0.304075	0.038664
379	-0.951607	0.764171	0.756629	-0.711175	-0.685909	4.108069	0.295557	-0.988509	0.774683	0.738730	0.304075	-0.495042
399	-0.177500	1.560729	0.171570	-0.711196	-0.644152	-0.221194	0.661659	-0.651578	-1.293143	-1.353674	0.304075	-0.390902
...	...	...	...	...	...	...	...	...	...	...	...	...
445419	-0.870918	1.545312	0.533020	-0.711000	-0.482173	-0.180980	0.758568	-0.597766	-0.111528	0.738730	0.304075	0.154787
445406	-1.048434	0.507216	0.710682	-0.711192	1.032077	3.372737	0.909316	1.307028	-1.293143	0.738730	0.304075	0.499407
445385	-1.243599	0.954317	1.194658	-0.711186	0.541093	2.591448	1.034939	1.960754	0.479279	0.738730	0.304075	-0.261278
445396	1.000051	0.460964	-0.701424	-0.711196	0.463086	3.981683	0.719087	-0.739196	-1.293143	0.738730	0.304075	-0.831548
445402	-1.175014	1.447669	0.147065	-0.673413	-0.014132	-0.102851	0.378109	-0.824434	0.479279	0.738730	0.304075	-0.180814

Modelo	Metodo Codo	Silueta	Boulding
K-Means	2	2	2
C-Means	2	2	2
GMM	2	2	2

Aprendizaje no-supervisada¶

Instrucciones¶

Objetivo de la tarea¶

Dataset:¶

Lista de librerias¶

Carga de datos¶

Paso 1: Selección, limpieza y preprocesamiento de features¶

Subconjunto de datos¶

Duplicados¶

Seleccion de features¶

Chequeo de datos faltantes¶

Correlacion entre features¶

Escalamiento de datos¶

Detección de outliers¶

Dataset final¶

Paso 2: Clustering de canciones¶

K-Means¶

Paso 3: Interpretación de los clusters obtenidos¶

Paso 4: Selección de playlist¶

BONUS: Conectarse a la API de Spotify¶

Obtener información de canciones¶

Grabar la playlist en mi cuenta¶

Modelo	n_clusters	silueta	davies_bouldin
K-Means	2	0.250665	1.551570
C-Means	2	0.247316	1.580740
GMM-diag	2	0.238371	1.587731

	variance	cumulative_variance	N_PCA
0	0.349398	0.349398	1
1	0.105933	0.455331	2
2	0.096997	0.552328	3
3	0.090348	0.642676	4
4	0.081498	0.724174	5
5	0.069014	0.793188	6
6	0.056463	0.849651	7
7	0.044237	0.893888	8
8	0.037231	0.931119	9
9	0.033441	0.964560	10
10	0.024764	0.989324	11
11	0.010676	1.000000	12

	PC	feature	loading	type
0	PC1	energy	-0.469967	best
1	PC2	mode	-0.766313	best
2	PC1	acousticness	0.447399	weak
3	PC1	danceability	-0.393230	weak
4	PC1	instrumentalness	0.364452	weak
5	PC1	liveness	-0.162428	weak
6	PC1	speechiness	-0.106605	weak
7	PC1	valence	-0.420971	weak
8	PC1	tempo	-0.195038	weak
9	PC2	key	0.621789	weak
10	PC1	time_signature	-0.103514	weak
11	PC1	duration_ms	0.158561	weak

	cluster	acousticness	danceability	energy	instrumentalness	liveness	speechiness	valence	tempo	key	mode	time_signature	duration_ms
0	0	0.245993	0.559010	0.706121	0.082332	0.264044	0.072672	0.558234	120.836740	5.563073	0.662844	3.951261	239157.172592
1	1	0.885446	0.303837	0.162496	0.537127	0.150072	0.043350	0.180506	102.763305	5.049426	0.651368	3.762577	332163.961165

artist_name	track_name
Los Tres Ases	Ofrenda
Los Autenticos Decadentes	El Dinero No Es Todo
Los Jaivas	Amores De Antes
Sr71	Right Now
Andres Calamaro	La Verdadera Libertad
Ben Harper	Like A King
The Offspring	All Along
Quilapayun	El Sabioloco
Limp Bizkit	Full Nelson
Alejandro Sanz	Dale Al Aire Con Juan Habichuela Y Ketama
Helloween	Escalation 666
Deep Purple	Lucille Live 1972
Therion	Flesh Of The Gods
Pyotr Ilyich Tchaikovsky	The Nutcracker Op 71 C Danse Russe Trepak Temp...
Ricky Martin	She Bangs English Version
Los Miserables	Quien Mato A Marilyn
Caetano Veloso	Zumbi
Diego Torres	La Ultima Noche
Caetano Veloso	Meu Rio
Franco De Vita	Te Amo En Vivo
Miguel Bose	Dulce Pesadilla
Helloween	I Live For Your Pain
Divididos	El Arriero Live
Los Terricolas	Te Juro Que Te Amo
Deftones	Teenager
Nightwish	Dead Boys Poem
Quilapayun	Tio Caiman
La Oreja De Van Gogh	Pop
Linkin Park	Cure For The Itch
Limp Bizkit	My Way
Acdc	Give It Up
U2	When I Look At The World
Tiesto	Airtight
Millencolin	Hellman
Sergio Mendes	Going Out Of My Head
Los Terricolas	Dos Cosas
Outkast	Stankonia Stanklove Feat Big Rube Sleepy Brown
Gondwana	Libros Sagrados
Limp Bizkit	Boiler
Alejandro Sanz	El Alma Al Aire Extended Remix
The Offspring	Special Delivery
Luis Fonsi	Mi Sueno
Tommy Guerrero	Soul Miner
Andres Calamaro	Papa Say

	artist_name	album_name	track_name	track_id	acousticness	danceability	energy	instrumentalness	liveness	speechiness	valence	tempo	key	mode	time_signature	duration_ms	cluster
0	2Pac	The Rose That Grew From Concrete	A River That Flows Forever	0FvMwjhIU2xsQUj6jKZWrU	0.608	0.506	0.413	0.000000	0.2140	0.2630	0.581	98.720	6.0	1.0	4.0	140373.0	0
1	2Pac	The Rose That Grew From Concrete	Can U C The Pride In The Panther Album Versio...	1tUtkxKxeDlyV8y5BPusyI	0.126	0.603	0.733	0.000008	0.0766	0.4270	0.489	82.902	8.0	1.0	4.0	176866.0	0
2	2Pac	The Rose That Grew From Concrete	Family Tree	1kxnbvzdnFW2uleKZx6gIC	0.433	0.758	0.542	0.000000	0.0857	0.0502	0.591	93.377	1.0	0.0	4.0	199800.0	0
3	2Pac	The Rose That Grew From Concrete	God	1NzCGSToYZti9uKubFHBxx	0.938	0.616	0.123	0.080600	0.1160	0.2250	0.433	76.863	6.0	0.0	3.0	47866.0	1
4	2Pac	The Rose That Grew From Concrete	If There Be Pain	0GmCjP5g3yMqqBWhDGjw87	0.181	0.614	0.559	0.000000	0.1630	0.0413	0.631	81.456	9.0	1.0	4.0	271693.0	0

	artist_name	album_name	track_name	track_id	acousticness	danceability	energy	instrumentalness	liveness	speechiness	valence	tempo	key	mode	time_signature	duration_ms
1432	Linkin Park	Hybrid Theory (Int'l Only DMD w/ Altered iLiner)	A Place For My Head	4vsIwS56G2Khi36iLoIZgH	0.0144	0.603	0.908	0.000000	0.671	0.1840	0.457	133.063	11.0	1.0	4.0	184640.0
1433	Linkin Park	Hybrid Theory (Int'l Only DMD w/ Altered iLiner)	By Myself	6CqMnBhJkUNzZqkDq2Dz3u	0.0171	0.544	0.935	0.003010	0.297	0.1730	0.339	103.026	6.0	0.0	4.0	189800.0
1434	Linkin Park	Hybrid Theory (Int'l Only DMD w/ Altered iLiner)	Crawling	0EorNL6AhJkrR2h4IltoaQ	0.0466	0.580	0.702	0.000003	0.536	0.0337	0.299	105.076	4.0	1.0	4.0	208960.0
1435	Linkin Park	Hybrid Theory (Int'l Only DMD w/ Altered iLiner)	Cure For The Itch	27P3adXGQqiHjq0Y0C2fS7	0.1630	0.750	0.677	0.050900	0.247	0.2210	0.851	99.990	8.0	1.0	4.0	157200.0
1436	Linkin Park	Hybrid Theory (Int'l Only DMD w/ Altered iLiner)	Forgotten	6Y7F8eKKHWC7no0Jn0pX1I	0.0132	0.615	0.947	0.000000	0.366	0.1100	0.498	108.193	8.0	0.0	4.0	194426.0

	artist_name	track_name
1624	Los Tres Ases	Ofrenda
1461	Los Autenticos Decadentes	El Dinero No Es Todo
1498	Los Jaivas	Amores De Antes
2610	Sr71	Right Now
108	Andres Calamaro	La Verdadera Libertad
463	Ben Harper	Like A King
2691	The Offspring	All Along
2259	Quilapayun	El Sabioloco
1420	Limp Bizkit	Full Nelson
36	Alejandro Sanz	Dale Al Aire Con Juan Habichuela Y Ketama
1065	Helloween	Escalation 666
749	Deep Purple	Lucille Live 1972
2718	Therion	Flesh Of The Gods
2188	Pyotr Ilyich Tchaikovsky	The Nutcracker Op 71 C Danse Russe Trepak Temp...
2371	Ricky Martin	She Bangs English Version
1528	Los Miserables	Quien Mato A Marilyn
563	Caetano Veloso	Zumbi
770	Diego Torres	La Ultima Noche
548	Caetano Veloso	Meu Rio
938	Franco De Vita	Te Amo En Vivo
1736	Miguel Bose	Dulce Pesadilla
1068	Helloween	I Live For Your Pain
793	Divididos	El Arriero Live
1594	Los Terricolas	Te Juro Que Te Amo
765	Deftones	Teenager
1860	Nightwish	Dead Boys Poem
2290	Quilapayun	Tio Caiman
1404	La Oreja De Van Gogh	Pop
1435	Linkin Park	Cure For The Itch
1427	Limp Bizkit	My Way
25	Acdc	Give It Up
2799	U2	When I Look At The World
2741	Tiesto	Airtight
1766	Millencolin	Hellman
2578	Sergio Mendes	Going Out Of My Head
1577	Los Terricolas	Dos Cosas
1909	Outkast	Stankonia Stanklove Feat Big Rube Sleepy Brown
977	Gondwana	Libros Sagrados
1419	Limp Bizkit	Boiler
38	Alejandro Sanz	El Alma Al Aire Extended Remix
2700	The Offspring	Special Delivery
1652	Luis Fonsi	Mi Sueno
2777	Tommy Guerrero	Soul Miner
137	Andres Calamaro	Papa Say

	danceability	energy	key	loudness	mode	speechiness	acousticness	instrumentalness	liveness	valence	...	type	id	uri	track_href	analysis_url	duration_ms	time_signature	artist_name	track_name	cluster
0	0.601	0.789	4	-6.822	0	0.0328	0.039500	0.81200	0.1140	0.315	...	audio_features	3nBGFgfRQ8ujSmu5cGlZIU	spotify:track:3nBGFgfRQ8ujSmu5cGlZIU	https://api.spotify.com/v1/tracks/3nBGFgfRQ8uj...	https://api.spotify.com/v1/audio-analysis/3nBG...	205926	4	Polyphia	Playing God	1
1	0.660	0.760	11	-5.488	0	0.0332	0.000601	0.31200	0.0949	0.359	...	audio_features	0YPuRrM2NwzdtuShUKkts6	spotify:track:0YPuRrM2NwzdtuShUKkts6	https://api.spotify.com/v1/tracks/0YPuRrM2Nwzd...	https://api.spotify.com/v1/audio-analysis/0YPu...	215999	4	Polyphia	G.O.A.T.	0
2	0.655	0.763	2	-5.202	0	0.0526	0.037400	0.67800	0.0703	0.445	...	audio_features	0Qj7PB41b6XgkApKPwDy1r	spotify:track:0Qj7PB41b6XgkApKPwDy1r	https://api.spotify.com/v1/tracks/0Qj7PB41b6Xg...	https://api.spotify.com/v1/audio-analysis/0Qj7...	350086	4	Polyphia	Ego Death (feat. Steve Vai)	1
3	0.681	0.827	6	-4.051	0	0.0651	0.055300	0.00000	0.2990	0.750	...	audio_features	7D4lgMbSs10AQ0zX3ZldJc	spotify:track:7D4lgMbSs10AQ0zX3ZldJc	https://api.spotify.com/v1/tracks/7D4lgMbSs10A...	https://api.spotify.com/v1/audio-analysis/7D4l...	152361	4	Polyphia	ABC (feat. Sophia Black)	0
4	0.597	0.845	1	-7.803	0	0.0335	0.002670	0.17300	0.2030	0.630	...	audio_features	3v3VFa7Dt32gNR27jfw7DG	spotify:track:3v3VFa7Dt32gNR27jfw7DG	https://api.spotify.com/v1/tracks/3v3VFa7Dt32g...	https://api.spotify.com/v1/audio-analysis/3v3V...	233819	4	Polyphia	40oz	0
5	0.461	0.962	8	-4.665	0	0.1200	0.000013	0.91500	0.3590	0.608	...	audio_features	6ZDHFz6h8d93EPAIq5hDuE	spotify:track:6ZDHFz6h8d93EPAIq5hDuE	https://api.spotify.com/v1/tracks/6ZDHFz6h8d93...	https://api.spotify.com/v1/audio-analysis/6ZDH...	265385	4	Polyphia	Champagne	0
6	0.655	0.763	2	-5.202	0	0.0526	0.037400	0.67800	0.0703	0.445	...	audio_features	7xyvWSVkxaoLdSuEdgdhBe	spotify:track:7xyvWSVkxaoLdSuEdgdhBe	https://api.spotify.com/v1/tracks/7xyvWSVkxaoL...	https://api.spotify.com/v1/audio-analysis/7xyv...	350086	4	Polyphia	Ego Death (feat. Steve Vai)	1
7	0.633	0.851	11	-5.724	0	0.0685	0.000569	0.89800	0.1400	0.592	...	audio_features	1U2yqd3Bp4vBuIQ3NGfS6C	spotify:track:1U2yqd3Bp4vBuIQ3NGfS6C	https://api.spotify.com/v1/tracks/1U2yqd3Bp4vB...	https://api.spotify.com/v1/audio-analysis/1U2y...	202163	4	Polyphia	O.D.	0
8	0.656	0.696	4	-5.935	0	0.0438	0.118000	0.45000	0.1610	0.400	...	audio_features	5l3WNlq4rag0fFP0TI9qeJ	spotify:track:5l3WNlq4rag0fFP0TI9qeJ	https://api.spotify.com/v1/tracks/5l3WNlq4rag0...	https://api.spotify.com/v1/audio-analysis/5l3W...	245779	4	Polyphia	The Worst	1
9	0.689	0.679	11	-5.239	0	0.0543	0.022300	0.71000	0.1030	0.922	...	audio_features	0uzwUl56ZPCJtRlqhG5OFo	spotify:track:0uzwUl56ZPCJtRlqhG5OFo	https://api.spotify.com/v1/tracks/0uzwUl56ZPCJ...	https://api.spotify.com/v1/audio-analysis/0uzw...	194590	4	Polyphia	Neurotica	0
10	0.601	0.789	4	-6.822	0	0.0328	0.039500	0.81200	0.1140	0.315	...	audio_features	6AhwAWzSlISc5ZvGonkgdN	spotify:track:6AhwAWzSlISc5ZvGonkgdN	https://api.spotify.com/v1/tracks/6AhwAWzSlISc...	https://api.spotify.com/v1/audio-analysis/6Ahw...	205926	4	Polyphia	Playing God	1
11	0.553	0.824	4	-5.261	0	0.0550	0.001850	0.82000	0.3050	0.608	...	audio_features	1W5zSvAeIXlw6odInPE4m5	spotify:track:1W5zSvAeIXlw6odInPE4m5	https://api.spotify.com/v1/tracks/1W5zSvAeIXlw...	https://api.spotify.com/v1/audio-analysis/1W5z...	242694	4	Polyphia	Reverie	0
12	0.664	0.835	8	-4.858	1	0.0442	0.035900	0.12800	0.1850	0.485	...	audio_features	5aVKIdM550lRzk7rFbPcF7	spotify:track:5aVKIdM550lRzk7rFbPcF7	https://api.spotify.com/v1/tracks/5aVKIdM550lR...	https://api.spotify.com/v1/audio-analysis/5aVK...	236653	4	Polyphia	Chimera (feat. Lil West)	0
13	0.421	0.931	0	-4.569	1	0.0397	0.000451	0.28700	0.0774	0.393	...	audio_features	0Jf3QiP8WrXEoFsaUsJi0b	spotify:track:0Jf3QiP8WrXEoFsaUsJi0b	https://api.spotify.com/v1/tracks/0Jf3QiP8WrXE...	https://api.spotify.com/v1/audio-analysis/0Jf3...	256333	4	Polyphia	Euphoria	0
14	0.640	0.846	9	-5.201	0	0.0298	0.021800	0.54100	0.1160	0.508	...	audio_features	2v7iJcMoQcN40fK9XEb42q	spotify:track:2v7iJcMoQcN40fK9XEb42q	https://api.spotify.com/v1/tracks/2v7iJcMoQcN4...	https://api.spotify.com/v1/audio-analysis/2v7i...	255455	4	Polyphia	Goose	0
15	0.600	0.817	4	-4.636	0	0.0488	0.000794	0.00321	0.4620	0.322	...	audio_features	2IMHE3XJcsqTIbSGOIY6Jy	spotify:track:2IMHE3XJcsqTIbSGOIY6Jy	https://api.spotify.com/v1/tracks/2IMHE3XJcsqT...	https://api.spotify.com/v1/audio-analysis/2IMH...	230951	4	Polyphia	Bloodbath (feat. Chino Moreno)	0
16	0.681	0.827	6	-4.051	0	0.0651	0.055300	0.00000	0.2990	0.750	...	audio_features	4c8TMfsKJIrRgCWs8LCEbQ	spotify:track:4c8TMfsKJIrRgCWs8LCEbQ	https://api.spotify.com/v1/tracks/4c8TMfsKJIrR...	https://api.spotify.com/v1/audio-analysis/4c8T...	152361	4	Polyphia	ABC (feat. Sophia Black)	0
17	0.553	0.885	0	-4.063	1	0.0385	0.000667	0.00072	0.2090	0.484	...	audio_features	6MYzjR2rH0hfz91FsaR1ox	spotify:track:6MYzjR2rH0hfz91FsaR1ox	https://api.spotify.com/v1/tracks/6MYzjR2rH0hf...	https://api.spotify.com/v1/audio-analysis/6MYz...	240006	4	Polyphia	So Strange (feat. Cuco)	0
18	0.364	0.423	2	-8.187	1	0.0343	0.001080	0.97500	0.6970	0.187	...	audio_features	4iDH45ZVIdHzDhLYd1FyKF	spotify:track:4iDH45ZVIdHzDhLYd1FyKF	https://api.spotify.com/v1/tracks/4iDH45ZVIdHz...	https://api.spotify.com/v1/audio-analysis/4iDH...	79688	3	Polyphia	All Falls Apart	1
19	0.674	0.712	0	-5.338	1	0.0888	0.093500	0.84700	0.1650	0.575	...	audio_features	71N78jUmfhLeRDlLd8elfl	spotify:track:71N78jUmfhLeRDlLd8elfl	https://api.spotify.com/v1/tracks/71N78jUmfhLe...	https://api.spotify.com/v1/audio-analysis/71N7...	194632	4	Polyphia	Genesis (feat. Brasstracks)	1